Google is deciding against using Samsung for its upcoming tensor chips.

 

Google set out on a goal to incorporate its own unique technology into Pixel smartphones, taking a cue from Apple.

Tensor, Google's first bespoke processor, debuted with the Pixel 6 series even though it was based on Exynos chips from Samsung made with a 5nm node manufacturing process. Samsung's semi-custom processor was still used in the Pixel 7 and will be in the future Pixel 8.

To introduce its completely customised Tensor G5 processor in 2025, Google appears to be intending to break off its relationship with the Korean firm and use TSMC (Taiwan Semiconductor Manufacturing Company).The Information cites a former Google official who claims that the original intention was to introduce a fully customised Tensor processor as early as the following year. The Redondo chip's manufacture, however, was delayed because it was difficult to get personnel on board and to coordinate development across numerous departments.

As previously stated, the chip's release has been postponed until 2025.

Another source echoes earlier claims, confirming that Google has actually chosen to switch the production of its Tensor G5 chip from Samsung to TSMC.The chip will utilise integrated fan-out technology and be built using TSMC's cutting-edge 3-nanometer manufacturing process, which will result in a smaller design and increased power efficiency.Samsung will continue to develop and produce the chipset for the upcoming Pixel 8 series, which will be released this year. Similar to this, the 2024 launch of Pixel smartphones will also have a Samsung processor, albeit with less Samsung components than prior incarnations.

Google hopes to achieve a higher level of hardware and software integration with a wholly bespoke processor, which will result in much better performance for their smartphones.

The user experience will be improved compared to the existing performance levels because to this tighter integration, which will enable optimised synergy between the hardware and software components.